Reservoir volume optim ization and performance evaluation of rooftop catchment systems in arid regions:A case study of Birjand,Iran

Faculty of Natural Resources and Environment,University of Birjand,Birjand 97175,Iran

Reservoir volume optim ization and performance evaluation of rooftop catchment systems in arid regions:A case study of Birjand,Iran

Zinat Komeh*,HadiMemarian,Seyed Mohammad Tajbakhsh

Faculty of Natural Resources and Environment,University of Birjand,Birjand 97175,Iran

Abstract

Thisstudy evaluated the performance of rooftop catchmentsystems in securing non-potablewater supply in Birjand,located in an arid area in southeastern Iran.The rooftop catchmentsystemsatseven study sites of different residential buildingswere simulated for dry,normal,and wet water years,using 31-year rainfall records.The trial and error approach andmass diagram method were employed to optim ize the volume of reservoirs in five differentoperation scenarios.Results showed that,during the dry water year from 2000 to 2001,for reservoirsw ith volumesof 200-20000 L,the proportion of days that could be secured fornon-portablewater supply wason average computed to be 16.4%-32.6%across all study sites.During the normal water year from 2009 to 2010 and the wetwater year from 1995 to 1996,for reservoirs w ith volumes of 200-20000 L,the proportions were 20.8%-69.6%and 26.8%-80.3%,respectively.Therefore,a rooftop catchment system showed a high potential tomeeta significantportion ofnon-potablewaterdemand in the Birjand climatic region.Reservoirvolumeoptimization using themass diagram method produced resultsconsistentw ith thoseobtained w ith the trialand errorapproach,exceptatsites#1,#2,and#5.At these sites,the trialand errorapproach performed better than themassdiagram method due to relatively highwater consumption.Itisconcluded that the rooftop catchmentsystem isapplicable under the same climatic conditionsas the study area,and itcan be used asa droughtmitigation strategy aswell. ©2017 Hohai University.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license(http:// creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords:Mass diagram analysis；Non-potablewater demand；Reservoir volume optim ization；Rooftop catchment；Rainwater harvesting

1.Introduction

Waterscarcity affectsmany placesaround theglobe.About 1.2 billion people,or almost one-fi fth of the world's population,live in areas of physical scarcity,and 500m illion people are approaching this situation.Another 1.6 billion people or almost one-quarter of the world's population face econom ic water shortage,without sufficient infrastructure to takewater from rivers and aquifers(Watkins et al.,2006).Urban developmentnotonly increases the frequency andmagnitude of the peak flow but also reduces the base flow.Therefore,water management in urban areas should seek to prevent negative hydrological changes(Price,2014).Rainwater catchmentsystems are one of themanagement and operationalmethods ofwaterharvesting thatcan affect the production of runoff and effectively improve the efficiency of rainwateruse in different land uses(Oweis,2001；Jacob,2007；Sturm et al.,2009). Rainwater catchment systems are known as compatible systems in terms of water supply in arid and sem i-arid areas and loss prevention in hum id regions(Vohland and Boubacar, 2009；Tubeileh et al.,2009).Rainwater harvesting in upstream areas reduces further pollution and the cost of treatment(Teemusk and Mander,2007).Rainwater harvesting for supplemental irrigation in dry seasons has been successfully used in many arid and sem i-arid regions(Richardson et al., 2004；Qiang et al.,2006；Short and Lantzke,2006；Arya and Yadav,2006).The possibility of developing the use of rainwater harvesting systems in cities w ith water scarcity was investigated by Zhang et al.(2010).In another study by Villarreal and Dixon(2005),rainwater collection systems forstoringwater in Dansen,Sweden were analyzed.They created a computationalmodel to determine the potential quantification of rainwater supply based on analysis of the optimal volume of a reservoir.Abu-Zreig et al.(2013)evaluated the potential of rainwater harvesting from rooftops in Jordan.In their study,it was estimated that about 14.7 m illion m3of rainwater could be extracted from rooftops,which would increase thewater budgetof thewhole country by 6%.Abdulla and A l-Shareef(2009)and Assayed etal.(2013)described the role of rooftop catchment systems in saving water in a large part of Jordan,as well.The effectiveness of the system for securing water in arid areas with low rainfall and irregular distribution of precipitation was also clarified by Yuan et al. (2003).The application efficiency of a rooftop catchment system in desert areas of Palestinew ith an average rainfall of 90 mm was confi rmed by Tabatabai Yazdi et al.(2006).The use of rooftop catchment systemswas also recommended for securing an emergency water supply in dry regions of Australia(Stanton,2005).

A rooftop catchment system can be divided into twomain parts:(1)an insulation surface that collects rainwater and(2) cisterns for rainwater storage on rainy days.The excess rainwater can be stored for use in dry seasons.In each rooftop catchment system,the cost of a water storage tank is the highest,so the determ ination of the optimal volume of reservoir is vital for structural stability and maximum capture of rainfall at am inimum cost(Kahinda et al.,2007；Tam et al., 2010).The reservoir size depends on runoff production potential and the demand for rainwater(Ghisi et al.,2007).Su et al.(2009)established a relationship between the volume of a reservoirand water supply for residentsof Taipei.Several other studieshave been conducted on hydrologicalanalysisof the performance of rooftop catchment systems for rainwater harvesting in residential areas.For instance,Jones and Hunt (2010)presented a computationalmodel to evaluate the performance of rainwater harvesting from rooftops in the Southern United States.Rahman et al.(2010)showed that larger rooftopsaremore efficient in termsofwater saving and financial benefi ts.Reservoir payback period analysis showed that the total costof reservoir construction can be amortized in 15-21 years,depending on the reservoir size and weather conditions.Im teaz et al.(2011)conducted a study to design rainwater collection tanks in Melbourne,Australia and to develop a comprehensive model for their performance analysis.Palla et al.(2011)studied the optimal performance of a rainwater harvesting system.They proposed a suitablemodel for assessment of the impacts of inflow,outflow,and storage volume alterations on system performance.In a survey by Campisano and Modica(2012),the optimal size of a reservoir was estimated based on daily rainfall,recorded at rainfall stationsnear the study region.Hashim etal.(2013)proposed a new way of designing a reservoir that could be implemented on a large scale,reducing the pressure on water resources. Singh et al.(2013)explored amethodology based on one of the most popular and versatile hydrologicalmodels,the soil conservation service-curve number(SCS-CN)model,and its variants.Resultsshowed that themodelhasan inherentability to incorporate the major factors of runoff production in rooftop/urban areas,i.e.,the surface characteristics,initial abstraction,and antecedent dry weather period(ADWP)of catchments.They concluded that the SCS-CNmodelwould be a better tool for runoff quantification than othermethodsonly using empirical runoff coefficients.

Birjand,in southeastern Iran,iscurrently experiencingwater shortage problems due to extensive development,drought conditions,and unmanaged agricultural activities in surrounding basins,creating a need to find alternative and potential sources of potable and non-potable water.Rooftop catchment systems can be an efficient source of water,especially in arid and sem i-arid areas.This study aimed at evaluating theperformanceof rooftop catchmentsystems in securing a non-potablewater supply in Birjand.The optimal volume of reservoirswas investigated based on comparison of the results from themass diagram method and trialand error approach.

2.M aterials and m ethods

2.1.Study area

Birjand is the capital of Southern Khorasan Province, centered at a latitude of 32°52′N and a longitude of 59°12′E (Fig.1),w ith an area of 42.7 km2and an elevation of 1491m above sea level.The weather in the area is categorized as a desert climate based on Gossen's classification approach (Memarian et al.,2016),w ith the xerotherm ic index of 321, and it normally has cold w inters and hot dry summers.The average annual precipitation is 159.7 mm,according to the daily rainfall data from 1970 to 2010,recorded at the Birjand Synoptic Station.According to the rainfall time series at the Birjand Synoptic Station,87%of the annual precipitation occurs from November to March(Fig.2).

2.2.Methodology

Fig.1.Geographic locations of study sites.

Fig.2.Averagemonthly rainfall during normal,dry,and wetwater years from 1970 to 2010 recorded at Birjand Synoptic Station (*means themonth of the next year).

Hydrological simulation of rooftop catchment systemswas conducted forseven residentialbuildings(Fig.1)w ith different numbersof unitsand a variety of rooftop areasand population densities(Table 1).An investigation was conducted to collect necessary information for rooftop catchmentsimulation during the summer of 2014.After questionnarieswere fi lled out,we were able to obtain the basic information about water consumption atdifferentsites in differentseasons(Table 1).

To simulate the performance of a reservoir,itwasnecessary to identify the variables that have an impact on the volume of reservoir.The evaporation componentwas ignored due to the consideration of covered reservoirs in hydrological analysis. The volume of runoffItin L at time stept,captured by a rooftop catchment system was estimated w ith Eq.(1):

whereφis the runoff coefficient,Rtis the daily rainfall inmm at time stept,andAis the rooftop area inm2.The volume of water storage was estimated using Eq.(2):

whereVtandVt-1are the volumes of collected water in L in the reservoir at time stepstandt-1,respectively；Ytis the volume of outflow in L for the non-potable water supply at time stept；andPtis thevolumeofoverflow in L at time stept. The volume of overflow is calculated w ith Eq.(3)(Villarreal and Dixon,2005；Ghisi et al.,2009；Hajani et al.,2013):

Table 1Characteristicsof study sitesand daily water consumption in differentseasons.

whereVmaxis themaximum storage capacity of the reservoir in L.

In consultation w ith local authorities,itwas suggested that we apply and introduce some executable optim ization approaches,which could be easily used by local experts.The massdiagram method and trialand errorapproach benefi t from these properties.Furthermore,these techniquesare executable through computer programs or can be optim ized using some othermathematicalalgorithms like linear programm ing.Thus, themass diagram method wasutilized to calculate the optimal volume of a reservoir.W ith thismethod,theoptimalvolumeof reservoirSin L isequal to themaximum vertical distance between the inflow mass curve(cumulative inflow)and outflow mass curve(cumulative demand)(Subramanya,2008),as expressed in Eq.(4):

whereVdis the volume of water demand in L,andVSis the volume of water inflow in L.The mass diagram method requires determ ining the critical period in time series w ith a maximum difference between cumulative inflow(rainfall)and cumulative outflow(water demand).According to Frasier and Myer(1983),average annual precipitation is not a good measure for determ ining the potential amount of extractable water from rooftop catchment systems.Therefore,it is essential to analyze rooftop catchment systems based on wet and dry spellsw ith specified return periods.In thisstudy,daily rainfall statistics at the Birjand Synoptic Station during the period from 1970 to 2010 were employed for hydrological analysis.Hydrological analysis involved computations of parametersIt,Vt,Pt,andS.Based on the comparison of annual precipitation records w ith the average annual precipitation (i.e.,159.7mm)over the period from 1970 to 2010,thewater years from 1995 to 1996,2000 to 2001,and 2009 to 2010were considered the wet(annual precipitation of 239.5 mm),dry (annual precipitation of 62.5 mm),and normal(annual precipitation of 166.3 mm)conditions,respectively,in hydrological analysis of rooftop catchment systems.As depicted in Fig.3,themonthsof November,December,January,February, and March were considered wet seasons,and others were considered dry seasons in the simulation.

Fig.3.Ombrotherm ic curves of Birjand Synoptic Station for period from 1970 to 2010(*means themonth of the next year).

As a general rule,in order to avoid setting the reservoir water levelatzero on the fi rstday of the study period,awarm-up year should be simulated atallstudy sites.However,due to the lim ited amount of storage in this climatic region,all storage is consumed by the end of the warm-up year.Therefore,at the start of the follow ing year,the water level in the reservoirwas set at zero.

At all study sites,rooftop material was made of water insulation and asphalt.With regard to the type andmaterialof rooftops,the runoff coefficient was determ ined to be 0.85 (Herrmann and Hasse,1997).The amount of water output required for non-potable water demand of residents was estimated based on the obtained information(through questionnaires)about the average water consumption of residents in yard cleaning,car washing,and green space irrigation.The domestic non-potable water demand considered in this study did not include water consumption in showers,flushes,and dish washing.Based on the information extracted from questionnaires,the coefficient of domestic non-potable water consumption,which is the percentageof the domestic non-potable water consumption accounting for the totalwaterconsumption, was determined to be 0.2 and 0.14 for dry and wet seasons, respectively,forvillas.However,thiscoefficient for thestudied apartmentswasestimated at0.07 forboth dry and wetseasons.

W ith regard to the rooftop area and numberof residents,the trialand error approachwasused to determ ine the appropriate capacity of reservoirs.The reservoir capacities of 200,500, 1000,2000,3500,5000,6500,8000,10000,15000,20000,and 30000 L(only for study site#7)were chosen for simulation. Two key factorswere considered in the selection of thevolume of reservoir.The fi rst factor is the space needed for placing reservoirs in the yard.For this factor,reservoirswith a volume greater than 20000 L cannot be placed in the yard of villas. However,for an apartment complex there ismore space that accommodates reservoirs w ith a volume up to 30000 L.The second factor is the cost of reservoir purchases and construction.Field surveys and econom ic analysis indicate thatbuilding a reservoirw ith a volume greater than 20000 L w ill cost about5000 dollars.Thiscostcould be economically justifiable for hum id areas or a country other than Iran.However,it should be considered that the study area hasa dry climatew ith a small volume of harvestable rainwater.Furthermore,in Iran, water is subsidized and its price is very low(about0.3 dollars per cubic meter),in comparison with other countries.Therefore,based on the calculations(Memarian et al.,2015),a rainwater harvesting system w ith expensive and bulky reservoirs and a service lift ofmore than 30 years does notmake econom ic sense.However,large reservoirs would be more econom ically justifiable for the apartment complex,as comparedw ith villas.Generally,itisconsidered that reservoirs w ith a volume less than 200 L cannotmeet the need of residents for rainwater harvesting and,consequently,are not econom ically justifiable in this area.

Five operation scenarios were defined to determine the optimal volume of reservoirs for securing water demand, especially in dry seasons.The base scenario S1 was defined based on the basic information aboutwater consumption inwet seasons obtained from questionnaires,as shown in Table 1. The operation scenarios S2,S3,S4,and S5 were defined, respectively,by reducing 10%,20%,30%,and 50%of water consumption in wet seasons.To define these operation scenarios,rainfall records in wet seasons during the period from 1970 to 2010 were investigated.Statistical records indicate a decrease of rainfall in wet seasons over a range from 10%to 50%in this period.By assuming the adaptation of water consumption to rainfall reduction inwetseasons,the operation scenarioswere defined asa droughtm itigation strategy.In fact, by reducing the water consumption in wet seasons,we can supply thewater needed in dry seasons asmuch as possible.

3.Results and discussion

Fig.4.Volumes ofwater storage in reservoirsw ith differentsizes for non-portablewater supply atdifferentstudy sites during dry,wet,and normal years.

Results from hydrologicalsimulationw ith themethodology described above show that,during the dry water year from 2000 to 2001(referred to as thedry yearhereafter),thevolume of rainwater harvested from rooftop catchment systems was m inimal,such that a reservoir w ith a volume of 1000 L(at study sites#1,#2,#4,#5,and#6)could store themaximum amountof extracted water(Fig.4(a)).However,during thewetwater year from 1995 to 1996(referred to as the wet year hereafter),the extractable amount of water was higher.With theoptim ization of the reservoirvolume,moredaysduring dry seasons are secured in terms of non-potable water demand of residents(Ghisietal.,2007).W ith regard to high precipitation during the wet water year,using larger reservoirs leads to maximum rainwater harvesting.When the volume of a reservoir increased to 20000 L,75%of non-potablewater demand was satisfied(Fig.4(b)).During the normal water year from 2009 to 2010(referred to as thenormalyearhereafter),using a reservoirw ith a volume of 15000 L,60%of non-potablewater demand was satisfied(Fig.4(c)).

Aspresented in Fig.4,during thedry,wet,and normalyears, theamountsof rainwater storage in reservoirsw ith volumesof 200-20000 L fornon-potablewatersupplywere2.8-16.0m3, 4.8-49.4m3,and 3.5-45.6m3,respectively.As confi rmed by Rahman etal.(2010),the amountof rainwater harvested from small rooftops is partial.Moreover,inmost cases,the amount of themaximum rainwater storage in large reservoirs is constant.This implies that,regarding the maximum potential amount of rainwater harvesting and daily non-potable water demand,using large reservoirshas littleeffectonwater supply, except for the casesatsite#7 during thewetand normalwater years.

Fig.5(a)presents the volumes of water overflow from reservoirs,which are 3.7-0.2m3on average from reservoirsw ith volumes of 200-2000 L during the dry year.Under the same conditions,the volume ofwater overflow from reservoirsw ith volumesgreater than 2000 L isestimated to bezero.During the wetand normal years,the average volumes ofwater overflow from reservoirs w ith volumes of 200-20000 L are 22.0-2.2m3and 15.0-0.7m3,respectively(Fig.5(b)and(c)).

Tables2 through 4 show the proportion of days that can be secured for non-potablewater supply during the dry,normal, and wet years.During the dry year,the proportion for reservoirs w ith volumes of 200-2000 L is 16.4%-32.2%, respectively,and the proportion for reservoirsw ith volumesof 2000-20000 L is 32.6%on average.During the normal and wet years,for reservoirs with volumes of 200-20000 L, 20.8%-69.6%and 26.8%-80.3%of days can be secured for non-potable water supply,respectively.These values are averages across all study sites.

According to Tables 2 through 4,during the dry year, reservoirsw ith volumesof500-3500 L can reach thehighest amounts of water storage.However,due to dry conditions, reservoirsw ith a volume greater than 3500 L cannot receive and store a greater volume ofwater.During thewetyear,the highest storage is achieved by reservoirs w ith volumes of 8000-30000 L.

To determ ine the optimal volume of a reservoir,themass diagram method was also employed.With this method,the optimal volume was determined based on themaximum difference between input and demand curves,which was influenced by factors such as the building area,number of residents,and volume of non-potable water consumption. Fig.6 depicts themass diagrams for different sites during the normal and wet years in the base scenario S1.

Tables 5 and 6 present the optimal volumes of the reservoir of the study sites obtained w ith the mass diagram method during the normal and wet years in different scenarios(S1 through S5),and the comparisons of the optimalvolume of the reservoir in the base scenario S1 w ith those in other scenarios(S2 through S5).

Fig.5.Volumesofwateroverflow from reservoirswith differentsizes for different study sites during dry,wet,and normal years.

Table 2Proportions of days that can be secured for non-potable water supply at different study sites during dry year.

Table 3Proportions of days that can be secured for non-potable water supply at different study sites during normal year.

Table 4Proportions of days that can be secured for non-potable water supply at different study sites during wet year.

As shown in Fig.6,w ith a decrease in water consumption in thewetseason,asdefined in differentscenarios,the slope of the demand curve decreases.With respect to a constant rate of input,the distance between the input and demand curves increases,exceptat sites#1,#2,and#5.As a result,in order to preventoverflow,the reservoir volume should be considerably increased.In these buildings,the demand curve is above the inputcurve on thewhole.Therefore,w ith a reduction ofwater consumption in thewet season,the distance between the two curves decreases,and,accordingly,the reservoir volume is reduced.However,in the case of site#1,the trend of changes in the optimal volume of the reservoir in different scenarios during thewet year is contrary to that during the normal year (Tables5 and 6).This occurs because the input curve is above the demand curve during thewetyear.During the normaland wet years,w ith a reduction of water consumption in the wet season,site#5 shows the greatest reduction in the optimal volume of the reservoir in different scenarios,compared to that in the base scenario.During the normal year,w ith a reduction ofwater consumption in wetseasons,site#7 shows a significant increase in the optimal volume of the reservoir, i.e.,46.4%in S5,compared to that in the base scenario.This occurs because of the decrease in the slope of the demand curve,resulting in an increase in the distance between the input and demand curves.During thewet year,site#1 shows the same condition,in which the optimal volume of the reservoir is increased by 43.28%in S5,compared to that in the base scenario.

As shown in Fig.5 and Tables 2 through 4,based on the trial and error approach,the optimal volume of the reservoir was determ ined based on two criteria:them inimum volume of overflow obtained w ith Eq.(3)and maximum number of days that can be secured for non-potable water supply. Comparison of the optimal volumes obtained from themass diagram method and trial and error approach shows anR2of 0.92 and 0.91 in normaland wet years,respectively.Thus,it can be seen that reservoir volume optim ization using the mass diagram method leads to results consistentw ith those obtained from the trialand error approach,exceptatsites#1,#2,and#5(Table 7).These buildings,w ith rooftop areas of 151,80,and 70 m2,have six,seven,and eight residents, respectively.Their water consumption is higher relative to the catchmentarea.At these sites,themass diagram method overestimates theoptimalvolumeof the reservoir.Hence,the trial and error approach is preferable.

Fig.6.Mass diagrams for different study sites during normal and wet years in base scenario.

Table 5Optimal volumes of reservoir of study sites in differentscenarios obtained w ithmass diagram method during normal year and comparison of optimal volume of reservoir in S1 w ith those in other scenarios.

Table 6Optimal volumes of reservoirs of study sites in different scenarios obtained w ith mass diagram method during wet year and comparison of optimal volume of reservoir in S1 w ith those in other scenarios.

Overall,results indicate the applicability of a rooftop catchment system to securing domestic non-potable water supply in Birjand,especially in dry conditions.Calculated results based on the standardized precipitation index(SPI)in Birjand over the period from 1970-2010 indicate that the water years of 1972-1973 and 2000-2001 demonstrated dry conditions,and thewater year of 1984-1985 was categorized as severe drought.Meanwhile,thewater years of 1973-1974, 1981-1982,1985-1986,1990-1991,1995-1996,and 2008-2009 represented wet conditions.In 1998,Birjand entered a long period of drought that has remained up to the present(Memarian et al.,2016).Therefore,a rooftop catchment system can be used as a droughtm itigation strategy in the study area.This is supported by severalworks in arid and sem i-arid regions around the globe(Stanton,2005；Abdulla and Al-Shareef,2009；Assayed et al.,2013；Abu-Zreig et al.,2013).As shown in this study,the use of a rooftop catchment system can reduce the pressure on water resources and consequently decrease the cost of household water.For example,atsites#4 and#6,w ith total rooftop areasof 85 and 90m2,the total non-potablewater demand during the normal and wetyears can be secured using reservoirsw ith volumesof 15000 L and 8000 L,respectively.At site#7,using reservoirs w ith volumes of 20000 L and 30000 L,68.8%and 80.3%of the total non-potable water demand can be secured,respectively,during the wet year.

Table 7Comparison of optimalvolumesof reservoiratstudy sites in S1 obtained w ith mass diagram method and trial and error approach in wet and normal years.

4.Conclusions

This study addressed the follow ing issues:

Reservoir volume optim ization using the mass diagram method produced outcomes consistent w ith those acquired through the trialand error approach,exceptatsites#1,#2,and #5.At these sites,the trial and error approach was preferable to themass diagram method.

During the dry year,for reservoirs with volumes of 200-20000 L,the proportion of days that can be secured for non-portable water supply was on average determ ined to be 16.4%-32.6%.During normal and wet years,for reservoirs w ith volumes of 200-20000 L,the proportions are 20.8%-69.6%and 26.8%-80.3%,respectively.

Comparison of scenarios S2 through S5 w ith the base scenario S1 under different climatic conditions showed significant differences in the optimal volume of reservoir.For example,during the normal year,w ith a reduction of water consumption in wet seasons,site#7 showed a significant increase in the optimal volume of the reservoir in S5,w ith an increasing rate of 46.4%as compared w ith that in S1.This would secure the non-potable water supply by as much as 46.4%ormore in the dry year.

In conclusion,the rooftop catchment system is applicable under the same climatic conditions as the study area and can secure a considerable portion of non-potablewater demand of residents.Therefore,with consideration of the existing water crisisinBirjand,rainwaterharvestingbasedon rooftop catchment systemscan beused asadroughtmitigation strategy in thisarea.

Acknow ledgements

The authors acknow ledge all residents who were questioned in this study,and the University of Birjand for providing necessary data.

Abdulla,F.A.,A l-Shareef,A.W.,2009.Roof rainwater harvesting system for household water supply in Jordan.Desalination 243(1-3),195-207. http://dx.doi.org/10.1016/j.desal.2008.05.013.

Abu-Zreig,M.,Hazaymeh,A.,Shatanaw i,M.,2013.Evaluation of residential rainfall harvesting systems in Jordan.Urban Water J.10(2),105-111. http://dx.doi.org/10.1080/1573062X.2012.709255.

Arya,S.L.,Yadav,R.P.,2006.Economic viability of rainwater harvesting by renovating village ponds in smallagriculturalwatershed of johranpur(HP). Agric.Econ.Res.Rev.19,71-82.

Assayed,A.,Hatokay,Z.,Al-zoubi,R.,Azzam,S.,Qbailat,M.,Al-Ulayyan,A.,Ma'ab,A.S.,Bushnaq,S.,Maroni,R.,2013.On site rainwater harvesting to achieve householdwater security among ruraland peri-urban communication in Jordan.Resour.Conserv.Recycl.73,72-77.http:// dx.doi.org/10.1016/j.resconrec.2013.01.010.

Campisano,A.,Modica,C.,2012.Optimal sizing of storage tanks for domestic rainwater harvesting in Sicily.Resour.Conserv.Recycl.63,9-16. http://dx.doi.org/10.1016/j.resconrec.2012.03.007.

Frasier,G.W.,Myer,L.E.,1983.Handbook ofWater Harvesting:Agricultural Handbook(No.600).Department of Agriculture,Agricultural Research Service,USA.

Ghisi,E.,Bressan,D.L.,Martini,M.,2007.Rainwater tank capacity and potential for potable water savings by using rainwater in the residential sector of southeastern Brazil.Build.Environ.42(4),1654-1666.http:// dx.doi.org/10.1016/j.buildenv.2006.02.007.

Ghisi,E.,Tavares,D.D.F.,Rocha,V.L.,2009.Rainwater harvesting in petrol stations in Brasilia:Potential for potable water savings and investment feasibility analysis.Resour.Conserv.Recycl.54(2),79-85.http:// dx.doi.org/10.1016/j.resconrec.2009.06.010.

Hajani,E.,Rahman,A.S.,A l-Amin,M.,Rahman,A.,2013.Reliability Analysis for rainwater harvesting system in peri-urban regions of Greater Sydney,Australia.In:20th International Congress on Modeling and Simulation,Adelaide,pp.2625-2631.

Hashim,H.,Hudzori,A.,Yusop,Z.,Ho,W.S.,2013.Simulation based programm ing for optim ization of large scale rainwater harvesting system: Malaysia case study.Resour.Conserv.Recycl.80(1),1-9.http:// dx.doi.org/10.1016/j.resconrec.2013.05.001.

Herrmann,T.,Hasse,K.,1997.Ways to get water:Rainwater utilization or long-distance water supply?A holistic assessment.Water Sci.Technol. 36(8-9),313-318.http://dx.doi.org/10.1016/S0273-1223(97)00580-5.

Im teaz,M.A.,Ahsan,A.,Naser,J.,Rahman,A.,2011.Reliability analysis of rainwater tanks in Melbourne using daily water balance model.Resour. Conserv.Recycl.56(1),80-86.http://dx.doi.org/10.1016/j.resconrec. 2011.09.008.

Jacob,M.,2007.The Eonomics of Rainwater Tanks and Alternative Water Supply Options:A Report Prepared for Nature Conservation Council of NSW.Australian Conservation Foundation and Environment Victoria. Marsden Jacob Associates,Melbourne.

Jones,M.P.,Hunt,W.F.,2010.Performanceof rainwater harvesting systems in the southeastern United States.Resour.Conserv.Recycl.54(10),623-629. http://dx.doi.org/10.1016/j.resconrec.2009.11.002.

Kahinda,J.,Taigbenu,A.,Boroto,J.,2007.Domestic rainwater harvesting to improve water supply in rural South A frica.Phy.Chem.Earth 32(15), 1050-1057.http://dx.doi.org/10.1016/j.pce.2007.07.007.

Memarian,H.,Tavasoli,A.,Tajbakhsh,M.,Komeh,Z.,Abbasi,A.K., Parsayi,L.,2015.Technical report:A guideline for designing and optimizing rainwater reservoirs in buildings(Case study:Golestan Province, Iran).Iran.J.Rainwater Catchment Syst.5(4),55-68(in Persian).

Memarian,H.,Pourreza Bilondi,M.,Rezaei,M.,2016.Drought prediction using co-active neuro-fuzzy inference system,validation,and uncertainty analysis(Case study:Birjand,Iran).Theor.Appl.Climatol.125(3), 541-554.http://dx.doi.org/10.1007/s00704-015-1532-9.

Oweis,T.,2001.Water Harvesting:Indigenous Know ledge for the Future of the Drier Environments.International Center for Agricultural Research in the Dry Areas,A leppo.

Palla,A.,Gnecco,I.,Lanza,L.G.,2011.Non-dimensional design parameters and performance assessment of rainwater harvesting systems.J.Hydrol. 401(1-2),65-76.http://dx.doi.org/10.1016/j.jhydrol.2011.02.009.

Price,K.,2014.Effects ofwatershed topography,soils,land use,and climate on base fl ow hydrology in hum id regions:A review.Prog.Phys.Geogr. 35(4),465-492.http://dx.doi.org/10.1177/0309133311402714.

Qiang,Z.,Yuanhong,L.,Manjin,C.,2006.Effect of low-rate irrigation with rainwaterharvesting system on the dry farming.In:Proceeding of the 2nd International RWHM Workshop.Beijing.

Rahman,A.,Dbais,J.,Im teaz,M.,2010.Sustainability of rainwaterharvesting systems in multistory residential buildings.Am.J.Eng.Appl.Sci.3(1), 73-82.http://dx.doi.org/10.3844/ajeassp.2010.73.82.

Richardson,L.,Hairsine,P.,Ellis,T.,2004.Water Farms:A Review of the Physical Aspects ofWater Harvesting and Runoff Enhancement in Rural Landscapes.Cooperative Research Centre for Catchment Hydrology, Canberra.

Short,R.,Lantzke,N.,2006.Increasing Runoff from Roaded Catchments by Chemical Application.Department of Agriculture and Food,Western Australia,Adelaide.

Singh,P.K.,Yaduvanshi,B.K.,Patel,S.,Ray,S.,2013.SCS-CN based quantification of potential of rooftop catchments and computation of ASRC for rainwater harvesting.Water Resour.Manag.27(7),2001-2012. http://dx.doi.org/10.1007/s11269-013-0267-6.

Stanton,D.,2005.Roaded Catchments to Improve Reliability of Farm Dams, Bulletin 4660.Department of Agriculture and Food,Western Australia, Perth.

Sturm,M.,Zimmermann,M.,Schu¨tz,K.,Urban,W.,Hartung,H.,2009. Rainwater harvesting as an alternative water resource in rural sites in central northern Nam ibia.Phys.Chem.Earth 34(13),776-785.http:// dx.doi.org/10.1016/j.pce.2009.07.004.

Su,M.D.,Line,C.H.,Chang,L.F.,Kang,J.L.,Lin,M.C.,2009.A probabilistic approach to rainwater harvesting systems design and evaluation.Resour. Conserv.Recycl.53(7),393-399.http://dx.doi.org/10.1016/j.resconrec. 2009.03.005.

Subramanya,K.,2008.Engineering Hydrology,third ed.Tata McGraw-Hill Publishing Company Lim ited,New Delhi,pp.166-174.

Tabatabai Yazdi,J.,Ghodsi,M.,Haghayeghi,S.,Rahnavard,M.,2006.Rainwaterharvesting asan approach for rainfallmanagementinarid regions.In: The2nd ConferenceonWater ResourceManagement,Isfahan(in Persian).

Tam,V.,Tam,L.,Zeng,S.,2010.Costeffectivenessand tradeoff on theuse of rainwater tank:An empirical study in Australian residential decisionmaking.Resour.Conserv.Recycl.54(3),178-186.http://dx.doi.org/ 10.1016/j.resconrec.2009.07.014.

Teemusk,A.,Mander,U.,2007.Rainwater runoff quantity and quality performance from a green roof:The effects of short-term events.Ecol.Eng. 30(3),271-277.http://dx.doi.org/10.1016/j.ecoleng.2007.01.009.

Tubeileh,A.,Bruggeman,A.,Turkelboom,F.,2009.Effectofwaterharvesting on grow th of young olive trees in degraded Syrian dryland.Environ.Dev. Sustain 11(5),1073-1090.http://dx.doi.org/10.1007/s10668-008-9167-y.

Villarreal,E.L.,Dixon,A.,2005.Analysisof a rainwater collection system for domestic water supply in Ringdansen,Norrkoping,Sweden.Build.Environ.40(9),1174-1184.http://dx.doi.org/10.1016/j.buildenv.2004.10.018.

Vohland,K.,Boubacar,B.,2009.A review ofinsitu rainwaterharvesting(RWH) practicesmodifying landscape functionsin A fricandrylands.Agric.Ecosyst. Environ.131(3),119-127.http://dx.doi.org/10.1016/j.agee.2009.01.010.

Watkins,K.,Carvajal,L.,Coppard,D.,Fuentes,R.,Ghosh,A., Giamberardini,C.,Johansson,C.,Seck,P.,Ugaz,C.,2006.Human Development Report 2006-Beyond Scarcity:Power,Poverty,and the GlobalWater Crisis.United Nations Development Programme(UNDP). Palgrave Macmillan,New York.

Yuan,T.,Li,F.M.,Liu,P.H.,2003.Economic analysisof rainwater harvesting and irrigationmethods,with an example from China.Agric.WaterManag. 60(3),217-226.http://dx.doi.org/10.1016/S0378-3774(02)00171-3.

Zhang,M.,Chen,H.,Wang,J.,Pan,G.,2010.Rainwater utilization and storm pollution control based on urban runoff characterization.J.Environ.Sci. 22(1),40-46.http://dx.doi.org/10.1016/S1001-0742(09)600723.

Received 9 August 2016；accepted 20 February 2017 Available online 1 June 2017

*Corresponding author.

E-mail address:z.komeh@gmail.com(Zinat Komeh).

Peer review under responsibility of Hohai University.

http://dx.doi.org/10.1016/j.wse.2017.05.003

1674-2370/©2017 Hohai University.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license(http:// creativecommons.org/licenses/by-nc-nd/4.0/).